The Initial Costs and Maintenance Costs of Protocols

Abstract

Software-engineering academics focussed for many years on the costs of developing the first version of a product, and ignored the costs of subsequent maintenance. We taught our students the ‘waterfall model’, and biased research towards the sort of tools and ideas that complemented it, such as formal methods. Meanwhile the economics of software had changed. Software is now so complex that the only way to build version N is to start with version N-1. Iterative development methodologies now rule, and the tools that real developers say have helped them most in the last fifteen years are not theorem provers, but automated regression-testing and bug-reporting systems. Nowadays, the maintenance is the product.

Security engineers have been falling into a similar trap. For years, we thought that the problem of authentication began and ended with trustworthy bootstrapping. Once Alice and Bob shared that elusive session key - and could prove mathematically that no-one else did - we could type up the research paper and head for the pub. Again, the real world has changed. Security maintainability is the elephant in the living room; people know there’s an awful problem but are generally too polite to mention it (especially as we don’t really know what to do with the beast). Vendors used to not care very much; after all, people replace their mobile phones every year, and their PCs every three to five years, so why not just wait for the vulnerable equipment to be thrown on the skip? With luck, vulnerability scares might even help stoke the upgrade cycle.